Jayden Reed · Data integrity
Try a case from my normalization exercise. See what the rules do when a record is clear, incomplete or contradictory.
Read the evaluation noteslength: 30 inThe source gives both a value and a unit. Convert inches to centimetres and retain the original entry.
Personal project · Python
I built a normalization exercise around 48 synthetic or public-spec cases and kept 16 aside during revision. A short README and Python runner document how to reproduce the checks and how the exercise handles incomplete or contradictory inputs.
I used 32 cases while revising the rules, then checked the held-out 16 separately. The set covers unit conversions, missing information and incompatible claims.
Harborfield Commerce · Supplier records
Support was seeing the same problems under different ticket names. I reviewed 2,400 records with support and engineering, grouped the errors into 12 patterns, and helped turn them into a validation checklist that three support teams adopted.
Dimensions without a unit. Keep the supplier’s original value, record where it came from, and request the missing unit before converting it. Flag conflicting dimensions for a reviewer.
I’ve spent five years investigating supplier-data problems with support and engineering. I’m comfortable in SQL and learning Python. My next step is to help design and evaluate the workflows I used to fix after the fact.
I’d like to talk about a failure your team sees often enough to write down and test. Bring a bounded workflow, a few examples and the person who has to review the result.